Introduction
Ecological intensification of aquaculture sector aims at improving productivity of the aquaculture farms while reducing environmental impact and ensuring fish welfare. In order to overcome the various challenges of this paradigm shift the GAIN H2020 project dedicated a non-negligible part of its work to the Precision Fish Farming (PFF) framework (Fore et. al., 2018), testing among other things some real-time biomass monitoring systems in farming conditions. One of the case studies was led in a rainbow trout farm in Preore (Trento, Northern Italy), where several cohorts of fish were monitored during several months using a weight monitoring system. Here we focus on the development of a reliable forecast system of population growth where model results and parameters are updated in real-time through the assimilation of the weight measurements. The data assimilation procedure improves then the short-term forecasts of the population distribution which is of huge interests for farm management.
The research leading to these results has received funding from the European Union’s H2020 Framework Programme for Research and Innovation, under Grant Agreement No. 773330.
Methodology
Data Assimilation (DA) algorithms are being currently used in many scientific fields, e.g. mechanics, oceanography, meteorology, as they allow one to combine model output and field data as long as they are collected, avoiding the full re-calibration of the model. The purpose of DA is two-fold: i) to correct the prediction of output variables, based on the information provided by new data; ii) to improve the estimation of the un-observed state variables, which may include also model parameters. Most DA procedures describe the system state variables as stochastic variables which distribution changes in time in accordance with the differential equations governing the system; the system measurements are crucial to reduce the uncertainty on this distribution and possibly correct its mean value.
In the last decade, several fish biomass monitoring system were designed and commercialized, mainly destinated to salmon production as a consequence of high turnovers, and thus investments margins of this market. To our knowledge, no similar system was designed for land-based flow-through farming, and we then decided to use the Biomass Daily (BD) commercialized by Vaki Ltd in a trout farm in Northern Italy. In parallel of the installation and test of the device, we build a growth model able to forecast growth performance of individuals, based on temperature and feed ration. Trying to best take benefit of both elements, we implemented a DA approach that integrates daily measurements of the population into the model, in order to improve the reliability of daily population weight distribution forecast. The individual-by-individual representation of the whole population was replaced by a cohort approach based on the super-individual concept (Scheffer et al., 1995). The initial weights of this super-individuals, and the number of fish each one of them represents, where defined from the initial distribution of the population measured by BD. Daily feed ration, food composition, and water temperature were daily downloaded from the farmer’s management system.
Results and discussion
Results show that the system allows to correctly predict daily fish weight distributions, providing thus an interesting additional level of information to the farmer that used until now only daily average weight predictions. On one side daily average weight is fully in line with the farmer’s predictions and with the BD measurements. On the other side, the daily weight distribution fits with the BD observations, and is in line with the intermittent samples made by the farmer. This method could be of high relevance for farmer’s days-to-day decision process as its results allow one to more accurately quantify some key parameters for farm management as feed quantity, oxygen supply, economical value of the biomass, but also the need for cohorts to be divided or moved to a larger raceway.
References
Føre M., K. Frank, T. Norton, E, Svendsen, J. A. Alfredsen, T. Dempster, H. Eguiraun, W. Watson, A. Stahl, L. M. Sunde, C. Schellewald, K. R. Skøien, M. O. Alver, D. Berckmans, 2018. Precision fish farming: A new framework to improve production in aquaculture. Biosystems engineering 173 (20 18): 176 -193.
Pasetto, D, Arenas‐Castro, S, Bustamante, J, et al. Integration of satellite remote sensing data in ecosystem modelling at local scales: Practices and trends. Methods Ecol Evol. 2018; 9: 1810– 1821.
Scheffer, M., Baveco, J.M., DeAngelis, D.L., Rose, K.A., van Nes, E.H., 1995. Super-individuals as a simple solution for modelling large populations on an individual basis. Ecological Modelling 80 (2-3): 161:170.